statsExpressions 2.1.2

This patch release contains documentation fixes only; there are no changes to
computed results.

• New Return value schema article documenting, for each exported function, the
  full set of returned tibble columns (including the preformatted expression
  and n.expression columns) and the role of the internal add_expression_col()
  engine. The shared @returns documentation has also been expanded to cover
  previously undocumented columns (e.g. bf10, log_e_bf10, prior.scale,
  group1/group2, p.value.adj).

• Corrected inaccurate or incomplete documentation:

  • meta_analysis() now shows the meta-analysis test table (instead of the
    correlation one), no longer advertises a non-parametric option, and
    documents type = "random".
  • contingency_table() documents that type = "bayes" ignores paired, that
    "nonparametric" and "robust" run the parametric tests, and which
    arguments apply only to Bayesian analyses.
  • conf.level = NULL is no longer described as skipping confidence intervals.
  • extract_stats_type() documents its case-sensitive matching, return value,
    and the stats_type_switch() alias.
  • add_expression_col(): corrected the digits default and the
    descriptions of the *.text and prior.type arguments.
  • long_to_wide_converter() documents that, without subject.id,
    observations are always paired by row order.
  • pairwise_contingency_table() documents that ... is passed to
    fisher.test().
  • Fixed descriptions of the bugs_long, iris_long, and movies_long
    datasets.

• Added missing documentation for tidy_model_parameters() and
  tidy_model_expressions(), and the pairwise_comparisons() and
  pairwise_contingency_table() test tables in the README and the test details
  article.

• Raised minimum versions of several dependencies.